While most of the world’s AI-and-jobs debate stays at the level of forecasts, China has moved to policy. The General Office of the State Council has issued an 18-point set of guidelines on adapting employment policy to artificial intelligence, structured around six areas: creating jobs through AI, making AI adoption more employment-friendly, improving workers’ digital skills, strengthening employment services, protecting labour rights, and monitoring employment risks. Xinhua published the directive on October 23, and international coverage followed this week — Business Standard summarised the plan’s contents on November 2, citing reporting from Reuters.
The timing is not random. China passed 700 million generative AI users in the first half of 2026 according to official figures, and AI adoption is spreading from consumer chatbots into factories, offices and administrative work. The guidelines are the government’s attempt to steer what happens to the workforce on the other side of that curve — and the direction is notable for how much of it is about protection rather than acceleration.
The measures inside the plan
The most quoted single item is the “unemployment buffer period”: companies introducing AI at scale are encouraged to give affected employees advance notice and provide retraining, transfers to other roles, and help finding alternative employment before displacement takes effect. As Caixin Global framed it in its November 2 report, the policy seeks to shift from responding to layoffs after the fact toward encouraging AI to create jobs while limiting displacement.
Other measures, per the Business Standard summary of the Reuters reporting:
- Employment-risk assessments before AI projects are approved in sectors subject to market-access requirements, when deployment could significantly affect jobs.
- Cautious adoption in labour-intensive industries, with authorities asked to weigh the scale, intensity and pace of automation; AI is to be prioritised first in jobs with labour shortages or hazardous conditions, where it improves safety.
- Retraining mandates for institutions: universities and vocational colleges are encouraged to update courses, add AI specialisations, and build training programmes with businesses. Specialised AI training is proposed for graduates, young people, migrant workers and the unemployed.
- Labour-rights language for human-machine workplaces: employers are called on to protect wages, rest days, leave and workplace safety, while the government plans to regulate AI use in HR — targeting algorithmic discrimination, fraudulent recruitment and privacy violations.
- Income-distribution adjustments so workers share in AI’s economic gains, plus wider social-security coverage for people in flexible and gig-style employment.
- Monitoring and early-warning systems for AI-related employment risk, with industry and employment data shared across departments and contingencies folded into existing large-scale-unemployment response mechanisms.
What is genuinely new here
Much of this exists in fragments elsewhere — the EU AI Act touches some workplace uses, Singapore has run large-scale AI reskilling (we covered its 200-course push), and New Zealand’s own AI guidance remains largely principles-based. But China appears to be the first country to bundle job-creation, adoption rules, retraining, rights protection and risk monitoring into a single employment-policy document aimed specifically at AI. The buffer-period concept — notice and retraining before displacement, not benefits after — is the piece other governments considering AI-labour policy will study most closely, particularly its implied bargain: adoption is fine, but the pace of adoption carries conditions.
It also continues a thread we have tracked before. A Chinese court ruled last year that AI adoption cannot by itself justify dismissing a worker (we covered that decision here), and Beijing’s October employment package paired its 10.52 million new urban jobs with a formal AI-skills programme (our earlier analysis). The 18-point guidelines pull those instincts into one framework. Researchers outside China will now watch whether implementation matches the language — guidelines of this type set direction for ministries and provincial governments rather than binding companies, and enforcement is always the open question.
The read for New Zealand
NZ’s exposure to China’s model is indirect but real. A large share of this country’s exports — dairy, meat, timber, horticulture — flow into a Chinese economy that is explicitly trying to manage AI-driven workforce transition rather than let it run. Labour-market stability in China is, among other things, a demand signal for New Zealand inc. Closer to home, the guidelines sketch a toolkit that any small open economy faces versions of the same questions about: how to sequence AI adoption in exposed sectors, who pays for retraining, and what notice workers get. Gartner’s research team found last month that about 30 percent of workers replaced by AI are hired back within three years — evidence that companies themselves struggle to judge when replacement pays. A policy framework that prices caution into the adoption decision is at minimum a test case worth watching.
FAQ
What is China’s 18-point AI employment plan? A set of guidelines issued by the General Office of China’s State Council, published October 23, 2026, laying out 18 tasks across six areas — job creation through AI, employment-friendly adoption, digital-skills training, employment services, labour-rights protection, and employment-risk monitoring.
What is the “unemployment buffer period”? A mechanism the guidelines encourage companies to establish when rolling out AI at scale: employees get advance notice plus retraining, transfers or job-search support before displacement happens, rather than relying on unemployment insurance afterwards.
Do the guidelines ban AI-driven layoffs? No. The text as reported by Reuters and Xinhua encourages caution and support, and asks for risk assessments in some sectors, but does not prohibit companies from adopting AI or restructuring. Chinese law still governs dismissals separately.
Does New Zealand have an equivalent policy? Not as a bundled framework. NZ has AI guidance and skills initiatives, but nothing combining employment-risk assessment, mandatory-style notice expectations, and AI-specific retraining programmes in a single employment-policy document.
— CJ Murden, editor of Singularity.Kiwi. Former digital technologies teacher, author of AI-focused books. Writing with a New Zealand focus.
